OUP user menu

The role of cAMP- and cGMP-dependent protein kinases in the cardiac actions of the new calcium sensitizer, levosimendan

Heimo Haikala, Petri Kaheinen, Jouko Levijoki, Inge-Britt Lindén
DOI: http://dx.doi.org/10.1016/S0008-6363(97)00057-6 536-546 First published online: 1 June 1997

Abstract

Objective: The role of phosphodiesterase III inhibition and calcium sensitization in the cardiac actions of levosimendan, (R)-[[4-(1,4,5,6-tetrahydro-4-methyl-6-oxo-3-pyridazinyl)phenyl]hydrazono]propane dinitrile, was studied. Methods: Various heart preparations were used to investigate positive inotropy, chronotropy, coronary flow and calcium sensitivity of contractile proteins. The cAMP- and cGMP-dependent protein kinases (PKA and PKG) were inhibited by KT5720 and KT5823, respectively. Furthermore, the synthesis of cAMP was stimulated by forskolin and increased phosphorylation of troponin I was induced by isoprenaline. Results: In Langendorff guinea-pig heart, levosimendan (0.01–1 μM) and milrinone (0.1–10 μM) increased the left ventricular systolic peak pressure almost to the same extent. In the presence of KT5720 (1 μM) milrinone was devoid of positive inotropic activity. In contrast, KT5720 did not antagonize the inotropic effect of levosimendan at ≤0.03 μM (= up to the EC50 of levosimendan). The effects of levosimendan and milrinone on heart rate and coronary flow were not affected by KT5720. The PKG inhibitor, KT5823 (1 μM), on the other hand, potentiated the levosimendan-induced increase in coronary flow while it had no effect on the increase induced by milrinone. The mechanical parameters were not affected by KT5823. In the papillary muscle, the positive inotropic effect of milrinone but not that of levosimendan was potentiated by forskolin (0.1 μM). In contrast to milrinone, the positive inotropy by levosimendan was decreased by isoprenaline pretreatment (0.1 μM; 3 min). In line with this, the calcium-sensitizing effect of levosimendan was decreased in skinned fibers prepared from isoprenaline-treated hearts. Conclusions: Our results indicate that the cardiac effects of levosimendan at its therapeutically relevant concentrations were not mediated through PKA or PKG and its positive inotropy is therefore most probably due to the previously reported troponin-C-mediated calcium sensitization of contractile proteins.

Keywords
  • Calcium sensitization
  • Levosimendan
  • Protein kinases
  • KT5720
  • KT5823
  • Guinea-pig, heart

Time for primary review 26 days.

1 Introduction

Calcium sensitization of contractile proteins has become a new promising approach to increase contractile force in the failing heart. An advantage of this mechanism is that it does not increase intracellular calcium whereby arrhythmias due to spontaneous release of calcium from overloaded intracellular calcium stores may be avoided. Furthermore, since the amount of calcium required for the contraction force is lower during calcium sensitization, the energy needed for calcium handling is reduced. An important disadvantage of calcium sensitization may, however, be related to prolonged relaxation [1, 2]. In order to compensate the impaired relaxation, attempts to accelerate intracellular calcium handling by another mechanism have been included as a pharmacological property for some calcium sensitizers. For example, the calcium sensitizer, pimobendan, which increases the affinity of troponin C (cTnC) for calcium, accelerates calcium handling through phosphodiesterase III (PDE III) inhibition [3]. Pimobendan, however, is a very weak calcium-sensitizing agent [4, 5] and therefore in clinical use does not reach concentrations high enough to cause calcium sensitization. This makes it impossible to judge how successful the balance between the increased calcium affinity of cTnC and accelerated calcium handling would be in practice. In pimobendan therapy, the formation of the potent metabolite, UDCG 212Cl, further complicates the evaluation [5–7]. The calcium sensitizer, EMD 53998, acts beyond troponin in the contraction cascade [8, 9]. Thus, the target protein of EMD 53998 is not directly affected by calcium and therefore the compound prolongs relaxation despite its potent PDE III inhibitory property [2, 10].

The novel potent calcium sensitizer levosimendan, (R)-[[4-(1,4,5,6-tetrahydro-4-methyl-6-oxo-3-pyridazinyl)-phenyl]hydrazono]propane dinitrile, does not increase the affinity of cTnC for calcium [11, 12] but stabilises the calcium-induced conformation of this protein [13]. In other words, the binding of levosimendan itself to cTnC occurs in a calcium-dependent manner and therefore the calcium sensitization does not prolong relaxation [11, 14]. The presence of PDE III inhibition with subsequent accelerated calcium handling is therefore not necessary. However, in purified enzyme preparations levosimendan is a potent PDE III inhibitor and this makes it important to evaluate the role of PDE III inhibition in its cardiac actions.

The above examples of different types of calcium sensitizers show that the proportions of calcium sensitization and PDE III inhibition cannot be accurately evaluated just by analysing the relaxation of cardiac muscle and therefore the use of pharmacological tools is needed. The muscarinic receptor agonists, carbachol and adenosine, are the pharmacological tools most frequently used in the evaluation of PDE inhibition [5, 15, 16]. However, these agents may not only counteract the contribution of PDE inhibition to the positive inotropic effect of calcium sensitizers, but they may also eliminate the calcium-sensitizing mechanism as well by altering the phosphorylation state of contractile proteins. In order to avoid this kind of problem, we used the new specific inhibitors, KT5720 and KT5823, for the cAMP- and cGMP-dependent protein kinases (PKA and PKG), respectively [17, 18]. As far as we know, these protein kinase inhibitors have not previously been studied in isolated heart. In addition to these investigations, we carried out experiments with isolated papillary muscles and skinned fibers to further elucidate the ratio between calcium sensitization and PDE III inhibition in the positive inotropic effect of levosimendan.

2 Methods

2.1 Selection of the pharmacological tools

Our preliminary experiments in guinea-pig papillary muscles demonstrated that, depending on the concentration, carbachol may induce negative or positive inotropy, which supports the concept that it has several mechanisms of action [19–22]. At a seemingly ineffective concentration (≈10 μM) carbachol probably has balanced effects at least on two second-messenger systems which indirectly and fully counteract each other by properly changing the phosphorylation state of proteins. As evidence for the carbachol-induced alterations at the level of contractile proteins the compound has been found to increase the calcium sensitivity of myofibrils in intact muscle [15]. Furthermore, we found that carbachol abolished the stimulus strength sensitivity of twitch tension in electrically paced papillary muscles, which may be due to the increase in intracellular sodium concentration mediated through muscarinic receptor stimulation [23]. Since the positive inotropy by levosimendan was markedly dependent on the stimulus strength, we considered that the use of carbachol is questionable when characterising the role of PDE inhibition in levosimendan inotropy. The use of receptor agonists like carbachol and adenosine become even more questionable when we take into account that they may alter the concentration of cAMP in other compartments than those containing PDE III. This is probable since the PDE III isoenzyme may occur only in a particular form close to the sarcoplasmic reticulum [24]. Subsequently, the interaction of a calcium sensitizer with carbachol or adenosine reveals only how the different receptor-mediated second-messenger systems indirectly modify the response to the calcium sensitizer.

We have demonstrated in the present study that for levosimendan the above aspects were very important and they should not be overlooked. Therefore, the specific protein kinase inhibitors, KT5720 and KT5823, were used as pharmacological tools when the function of PKA or PKG needed to be antagonised in the isolated heart. Our aim was to achieve as complete inhibition of PKA or PKG as possible but still maintaining selectivity of inhibition. This requirement led us to choose 1 μM concentration for both KT5720 and KT5823. This concentration clearly exceeds the IC50 (0.06 μM) of KT5720 for PKA and the IC50 (0.234 μM) of KT5823 for PKG [17, 18]. KT5720 and KT5823 had no effects of their own in the present study and therefore we could not select the inhibitor concentrations by constructing concentration–response curves for them. For stimulation of the PKA-mediated biochemical reactions in papillary muscles we used a submaximal concentration (0.1 μM) of forskolin which increased cAMP synthesis by activating adenylate cyclase. Furthermore, isoprenaline (0.1 μM) and atenolol (1 μM) were used for stimulation and blockade of β-adrenoceptors, respectively. The PDE III inhibitor, milrinone, was a reference compound for levosimendan.

2.2 Preparations

2.2.1 Isolated heart

The guinea-pig (4 weeks old; weight 300–350 g) was sacrificed and the heart was rapidly excised and rinsed in ice-cold oxygenated perfusion buffer. A cannula was inserted into the aorta and retrograde perfusion of the heart began as soon as the heart was placed in a thermostatically controlled moist chamber of a Langendorff apparatus (Hugo Sachs Elektronik, KG, Germany). Modified Tyrode solution (37°C), gassed with 95% O2 and 5% CO2 in a thermostatically controlled bulb oxygenator, was used as a perfusion buffer. The composition of the Tyrode solution was (in mM) NaCl 135; MgCl2·6H2O 1; KCl 5; CaCl2·H2O 2; NaHCO3 15; Na2HPO4·H2O 1; glucose 10; pH 7.3–7.4.

The experiments were carried out under constant pressure conditions (50 mmHg). After a short pre-stabilization period (10 min) a latex balloon (size 4) was placed in the left ventricle. The balloon was attached to a stainless-steel cannula coupled to a pressure transducer. The latex balloon, the cannula and the chamber of the pressure transducer were filled with an ethylene glycol/water (1:1) mixture. The isovolumetric left ventricular pressure was recorded with a pressure transducer. The pressure signal was amplified by using a two-channel bridge amplifier (Type 301, Hugo Sachs Elektronik, KG, Germany) connected to a Macintosh IIci computer. At the beginning of the experiment, the volume of the balloon was adjusted to obtain a diastolic pressure of approximately 5 mmHg. Coronary flow (ml/min) was continuously recorded by an electromagnetic flow meter (Narcomatic RT-500, Hugo Sachs Elektronik, KG, Germany) with a flow probe inserted above the aortic cannula. The flow meter was also connected to the computer. Samples recorded for 3 s from the pressure and flow signals were digitised and collected into the computer at every 20 s. The beating rate (HR) and the left ventricular systolic pressure (LVSP) of the heart were obtained from the digitised (1 kHz frequency) pressure signals. The left ventricular pressure signals were used to calculate the maximum positive and negative pressure derivatives (+dP/dtmax and −dP/dtmax, respectively). The coronary flow was obtained from the digitised flow meter signal.

The hearts were allowed to stabilise approximately for 30 min to reach a steady state in the measured parameters. Thereafter, dimethyl sulfoxide (DMSO, 0.1% final concentration), the vehicle for the protein kinase inhibitors, KT5720 (1 μM) and KT5823 (1 μM), was added to the perfusion buffer. The heart was exposed to DMSO alone for 15 min before addition of the protein kinase inhibitor in order to check that DMSO was ineffective. The DMSO concentration was kept constant throughout the experiment. Levosimendan or milrinone was added to the buffer 60 min after the start of exposure of the heart to either of the protein kinase inhibitors or to DMSO. The concentration–response curves for 0.01–1 μM levosimendan and 0.1–10 μM milrinone were obtained in the absence and presence of the protein kinase inhibitors by increasing the concentrations at 15 min intervals.

2.2.2 Papillary muscle

The right ventricular papillary muscle of the guinea-pig heart was mounted for measurement of isometric tension as previously described by Haikala et al. [25]. For pacing of the papillary muscle the electrical pulses were conducted through platinum wire electrodes inserted on both sides of the muscle (1 mm apart from the muscle surface). The field stimulation occurred at 20% above threshold or at twofold threshold voltage. A force-displacement transducer was connected to a driver amplifier and a programmable scanner. The amplified signal was digitised at 1 kHz frequency by a programmable digitizer. Subsequently, twitch tension of the papillary muscle was analysed by a Macintosh IIci computer.

After a stabilising period of about 40 min various concentrations of levosimendan or milrinone (final concentrations 0.03, 0.1, 0.3, 1, 3, 10 and 30 μM) were successively added to the superfusion buffer at 10 min intervals. The same concentration range was used when levosimendan and milrinone were investigated during the accelerated synthesis of cAMP induced by forskolin (0.1 μM). Similarly, levosimendan was studied also in the presence of 1 μM atenolol. In another group of studies 1 μM levosimendan and 10 μM milrinone were tested before and after a short period (3 min) of exposure of the papillary muscle to 0.1 μM isoprenaline. In all studies data from 5–6 papillary muscles in each group were used to calculate the mean changes in twitch tension.

2.2.3 Skinned fibers

The calcium-sensitizing effect of levosimendan was investigated in two types of skinned fibers obtained either from unpretreated or isoprenaline-pretreated (0.1 μM for 10 min) isolated, spontaneously beating guinea-pig hearts. Fast and complete perforation of the cell membranes was achieved by retrogradely conducting saponin solution (250 μg/ml) through the aorta into the coronary arteries of the heart [14, 26, 27]. The fibers dissected from the papillary muscles were further mildly sonicated and treated with saponin for 30 min. The details of this procedure have been earlier described by Haikala et al. [14]. The slightly acidic pH of 6.7 was chosen in order to mimic the pH in the ischemic myocardium in which the calcium sensitivity is decreased and in which the benefits of levosimendan should be most prominent. The fibers were induced to contract in the desired free pCa (−log[Ca2+]) which was calculated by using a program developed by Fabiato and Fabiato [28]. The absolute stability constants used were as reported by Fabiato [29]. At the beginning of the experiment the fiber was stretched in ‘relaxing’ (calcium free) solution (22°C) until the resting tension amounted to approximately 2% of the maximum force produced by the fiber. Then, the ‘relaxing’ solution was replaced with the ‘activating’ (pCa 5.8) solution which roughly corresponds to the intracellular cytoplasmic calcium during the muscle contraction. When the tension produced by the fiber had reached a steady state, various concentrations of levosimendan (final concentrations 0.3, 1, 3, and 10 μM) were successively added to the solution at 6 min intervals. The maximum tension which was produced by the fiber at pCa 4.8 was determined after washing out the test compound.

2.3 Data analysis

In order to verify that the selected concentration range for levosimendan and milrinone produced positive inotropy to the same extent in the isolated heart the levosimendan group was compared with the milrinone group by using ANCOVA (repeated measures). The same statistical test was used when the hearts treated with the protein kinase inhibitors were compared with those exposed to levosimendan or milrinone alone. One-factor ANOVA (repeated measures) followed by Dunnett's t-test was used, when the measured parameters were compared to the initial values recorded just before the first concentration of levosimendan or milrinone was added to the test medium.

2.4 Drugs and reagents

Levosimendan and milrinone synthesised at Orion Pharma, as well as isoprenaline, KT5720 and KT5823 (Sigma Chemical Co., St Louis, USA) and forskolin (Calbiochem-Novobiochem, La Jolla, USA) were dissolved in DMSO. The amount of DMSO did not exceed 0.4% in any of the experiments and in isolated heart it was kept at 0.1%. All other chemicals were purchased from E. Merck (Darmstadt, Germany).

3 Results

3.1 Isolated heart

In Fig. 1, the differences of initials levels of HR, CF, LVSP and −dP/dtmax between various groups were not due to the effects of the protein kinase inhibitors, KT5720 and KT5823, since these differences were already present before the inhibitors were added to the perfusion buffer. Therefore, KT5720 and KT5823 at 1 μM were considered to be devoid of own effects.

Fig. 1

Graphs showing the effects of levosimendan (squares) and milrinone (circles) on the various parameters (heart rate, A; coronary flow, B; systolic pressure, C; −dP/dtmax, D) describing the function of isolated guinea-pig heart in the presence (open symbols) and absence (black symbols) of the PKA inhibitor, KT5720 (1 μM), or in the presence (striped symbols) of the PKG inhibitor, KT5823 (1 μM). Given are the initial levels (symbols not connected to the curves) recorded from 5 hearts just before levosimendan or milrinone and the mean values±s.e.m. (depicted only in one direction) for each recorded parameter in the presence of various concentrations of the test compounds (levosimendan and milrinone). The statistically significant interactions (P<0.05; ANCOVA, repeated measures) of levosimendan with KT5720 were seen on systolic pressure and −dP/dtmax and with KT5823 on coronary flow. Furthermore, the interaction of milrinone with KT5720 on systolic pressure and −dP/dtmax were found to be statistically significant. The asterisks denote the differences compared to the initial level (*P<0.05 and **P<0.01; repeated measures ANOVA followed by Dunnett's t-test).

Levosimendan concentration-dependently increased the HR to the same extent in the absence and presence of the PKA inhibitor, KT5720 (1 μM), or the PKG inhibitor, KT5823 (1 μM) (Fig. 1A). In all groups the maximum increase of 27–30% was seen at 1 μM levosimendan. The magnitude of the positive chronotropic effect of milrinone (0.1–10 μM) was comparable to that of levosimendan (0.01–1 μM), but unlike with levosimendan the maximum effect was not reached at the concentrations used. Furthermore, the PKA and PKG inhibitors did not change the effect of milrinone on the spontaneous HR (Fig. 1A).

Levosimendan alone increased the coronary flow concentration-dependently and maximally by 64% at 1 μM (Fig. 1B). The increase in coronary flow by milrinone alone did not differ from that induced by levosimendan except that the maximum effect was not reached at the milrinone concentrations used. The response to levosimendan or milrinone was slightly and to the same extent decreased in the presence of KT5720 (1 μM). In contrast, the levosimendan-induced increase in coronary flow was almost twofold in the presence of KT5823 (1 μM) (Fig. 1B). The PKG inhibitor, KT5823, did not alter the milrinone-induced increase in coronary flow (Fig. 1B).

The increasing effects of levosimendan alone (0.01–1 μM) and milrinone alone (0.1–10 μM) on the LVSP did not differ statistically significantly (P=0.356) from each other (Fig. 1C). The EC50 concentration for levosimendan was about 0.03 μM and that for milrinone was about 0.6 μM (Fig. 1C). Levosimendan increased LVSP maximally by 17 mmHg (+19%) at 0.3 μM and milrinone maximally by 14 mmHg (+15%) at 3 μM. In the presence of the PKA inhibitor, KT5720 (1 μM), the effect of levosimendan was not altered at concentrations up to its EC50 concentration, but it was reduced by 30% at 0.3 μM. In contrast, milrinone did not increase statistically significantly LVSP at any of the tested concentrations when it was combined with KT5720 (Fig. 1C). In the presence of the PKG inhibitor, KT5823, the increasing effect of levosimendan or milrinone on LVSP was similar to that when the compounds were tested alone (Fig. 1C).

The increasing effects of levosimendan and milrinone on +dP/dtmax were qualitatively similar to those on LVSP and therefore the data are not presented. Levosimendan and milrinone alone increased −dP/dtmax maximally by 37 and 33%, respectively (Fig. 1D). KT5720 was unable to decrease statistically significantly the effect of levosimendan on −dP/dtmax at the levosimendan concentrations giving up to 70% of the maximum effect. Only at the two highest concentrations was 19–31% antagonism seen (Fig. 1D). In contrast, the corresponding effect of milrinone was markedly antagonised by KT5720 at all effective concentrations of milrinone. The magnitude of the antagonism was 40 and 50% at 1 and 10 μM milrinone, respectively (Fig. 1D). The PKG inhibitor, KT5823, did not affect statistically significantly the increasing effect of levosimendan or milrinone on −dP/dtmax (Fig. 1D).

3.2 Papillary muscle

The concentration–response curve of levosimendan on twitch tension displayed a bell-shaped form. Furthermore, the levosimendan-induced increase in twitch tension markedly depended on the stimulus strength used in the pacing of the papillary muscle. A similar phenomenon was not seen with milrinone (Fig. 2). The EC50 concentrations for levosimendan were 0.1 and 0.3 μM when the papillary muscles were paced at twofold or at 20% above threshold voltage, respectively. The maximum increases of 330 mg (+127%) and 90 mg (+45%) in twitch tension were reached at 1 and 3 μM, respectively. The EC50 for milrinone was about 2 μM independently of the stimulus strength used. The maximum increase of 240 mg in twitch tension induced by milrinone was also not affected by the stimulus strength (Fig. 2).

Fig. 2

The effects of levosimendan (A) and milrinone (B) on twitch tension in isolated guinea-pig papillary muscles. The experiments were carried out by using stimulus strength at twofold threshold (filled circles) and at 20% above threshold (filled squares). Given are changes from initial level and each value represents the mean±s.e.m. of 5–6 muscles. The twitch tensions just before levosimendan or milrinone are presented in the attached tables.

In papillary muscles paced at twofold threshold voltage forskolin (0.1 μM) increased twitch tension by 27 and 23% when it was used to accelerate the cAMP synthesis in the levosimendan and milrinone studies, respectively (Fig. 3A,B). The levosimendan-induced increase in twitch tension was not markedly changed in the presence of forskolin (Fig. 3A). In contrast, the positive inotropic effect of milrinone was almost twofold when it was combined with forskolin (Fig. 3B).

Fig. 3

The interaction of 0.1 μM forskolin (filled squares) with levosimendan (A) or milrinone (B) on twitch tension in isolated guinea-pig papillary muscles. The pacing of the muscles was performed at twofold threshold voltage. Given are changes from initial level and each value represents the mean±s.e.m. of 5–6 muscles. The twitch tensions just before levosimendan or milrinone are presented in the attached tables. The twitch tension before forskolin is given in brackets.

The initial twitch tension changed less than 10% by treatment of the β1-adrenoceptor antagonist, atenolol (1 μM). However, atenolol abolished the stimulus sensitivity of the twitch tension and thereby eliminated the stimulus strength dependency of the inotropic response to levosimendan (Fig. 4).

Fig. 4

Interaction of 1 μM atenolol (open symbols) with levosimendan in guinea-pig papillary muscles paced at low (20% above threshold voltage; squares) or high (twofold threshold voltage; circles) stimulus strength. Given are changes from initial level and each value represents the mean±s.e.m. of 5–6 muscles. The twitch tensions just before levosimendan are presented in the attached tables. The twitch tension before atenolol is given in brackets.

The single 1 μM concentration of levosimendan increased twitch tension by 400 mg (Fig. 5A, first exposure; stimulus strength twofold threshold) whereas under similar experimental conditions the stepwise increase in levosimendan concentration from 0.03 up to 1 μM increased it by 330 mg (Fig. 2A, stimulus strength twofold threshold). The single 10 μM concentration of milrinone and its stepwise concentration increase from 0.03 up to 10 μM increased twitch tension by 379 and 208 mg, respectively (Fig. 2B and 5A).

Fig. 5

Graph showing the effect of levosimendan or milrinone on twitch tension in isolated guinea-pig papillary muscle before and after isoprenaline (Ispr) pretreatment. The run of experiments is from left to right. The striped and filled bars refer to the levosimendan and milrinone groups, respectively. Panel A shows the recorded twitch tensions (±s.e.m.) in control situation (C1), at 15 min after first exposure to 1 μM levosimendan (Ls1) or 10 μM milrinone (Mil1), at 30 min after wash-out of these compounds (W), at 3 min after exposure to 0.1 μM isoprenaline (Ispr), at 10 min after wash-out of isoprenaline (C2 = control 2), and at 15 min after the second exposure to levosimendan (Ls2) or milrinone (Mil2). The levosimendan- or milrinone-induced percent changes in twitch tension from the control levels before and after isoprenaline pretreatment (C1 and C2, respectively) are also presented (B). Each value represents the mean±s.e.m. of 6 muscles.

After the single exposure of papillary muscles to either 1 μM levosimendan or 10 μM milrinone and the subsequent 30 min wash-out period the twitch tension in the milrinone group but not in the levosimendan group decreased below the initial twitch tension (Fig. 5A, W versus C1, P<0.01). A similar phenomenon was seen also in the levosimendan group after subsequent exposure of the same papillary muscles to 0.1 μM isoprenaline for 3 min followed by 10 min wash-out (Fig. 5A, C2 versus W). Levosimendan at 1 μM and milrinone at 10 μM increased twitch to the same extent (from +180 to +190%) when these concentrations were tested before isoprenaline. After isoprenaline (and its wash-out period) milrinone was percentually as efficient as before isoprenaline, but the levosimendan-induced increase in twitch tension was only 78% (Fig. 5B).

3.3 Skinned fibers

In skinned fibers obtained from naive guinea-pig hearts levosimendan (0.3–10 μM) concentration-dependently increased the calcium-induced tension at pCa 5.8 from 7 to 26% of the maximum tension produced by the fibers at pCa 4.8 (Fig. 6). When the fibers were dissected from isoprenaline-pretreated (0.1 μM, for 10 min) hearts the levosimendan-induced increase in tension was only from 6 to 16% of the maximum tension.

Fig. 6

Graph showing the concentration-dependent effect of levosimendan on the calcium-induced (pCa 5.8) tension in skinned fibers obtained from unpretreated (filled bars) and isoprenaline-pretreated (0.1 μM, for 10 min; striped bars) guinea-pig hearts. Given are relative tensions expressed as percent of the maximum tension produced by the fiber at pCa 4.8. Each value represents the mean±s.e.m. of 7 fibers. *P<0.05, **P<0.01; compared with control.

4 Discussion

The novel calcium sensitizer, levosimendan ([1, 14], was found to be 10–20 times more potent as a positive inotropic compound than the PDE III inhibitor, milrinone. In the presence of the PKA inhibitor, KT5720 [18], milrinone was devoid of the positive inotropic effect, which verifies the concept that PDE III inhibition is its major mechanism of action [30]. In contrast, KT5720 did not antagonise the increase in LVSP produced by levosimendan at concentrations up to its EC50 (0.03 μM). In heart failure patients 95–98% of levosimendan is bound to plasma proteins and therefore its active free plasma concentration is less than 0.02 μM in therapy [31]. In the light of the present results the positive inotropy by levosimendan at its therapeutically relevant concentrations is not due to PDE III inhibition or to the activation of PKA. This perfectly agrees with the previously found lack of effect of 0.03 μM levosimendan on cAMP and phosphorylation of cTnI [11]. These parameters, however, increase at higher levosimendan concentrations [11]. In agreement, we found that at higher concentrations the levosimendan inotropy was partially antagonised by KT5720. In addition, the concentration–response curve of levosimendan on LVSP was bending downwards at high concentration range. This may be due to the levosimendan-induced opening of ATP-sensitive K+ channels which shortens the action potential, leading to reduced calcium influx and contraction force [32]. It seems that this mechanism of action of levosimendan became potentiated during PKA inhibition.

Some calcium-sensitizing compounds acting beyond troponin in the cascade of the contraction trigger seriously impair the relaxation of the cardiac muscle [1, 2]. In contrast, levosimendan which binds calcium-dependently to cTnC [13] is devoid of this disadvantage [14]. The present results show that even in the presence of a PKA inhibitor levosimendan does not impair relaxation but increases the negative dP/dtmax. The milrinone-induced increase in the negative dP/dtmax was markedly antagonised by KT5720. However, the effect was not completely abolished probably because of the increase in HR.

HR is known to affect the contraction force of the heart. In the present study, however, the levosimendan- and milrinone-induced increases in the spontaneous HR were similar in the presence and absence of KT5720. Therefore, the evaluation of the role of PKA in the inotropic effects of these test compounds was valid. The lack of effect of KT5720 on the positive chronotropy induced by the test compounds cannot be explained by the poor penetration of KT5720 into cardiac cells, since the positive inotropy by milrinone was completely abolished by KT5720. Therefore, it is probable that the positive chronotropy induced by levosimendan or milrinone is mediated through their direct or G-protein-mediated effects on ion channels producing an inward current in the atrial pacemaker cells.

KT5720 only tended to decrease the levosimendan- or milrinone-induced increase in coronary flow. Therefore, the differences seen in the positive inotropic effects of levosimendan and milrinone were not affected by changes in the coronary flow. These findings also suggest that the major mechanism of the coronary dilatation by milrinone is not mediated by the same PKA subtype as that involved in its positive inotropic effect.

The PKG inhibitor, KT5823 [17], did not antagonise the positive inotropic effect of levosimendan or milrinone. Also no interaction between KT5823 and the test compounds could be shown on the positive chronotropic effects. However, KT5823 potentiated the levosimendan- but not the milrinone-induced increase in coronary flow. Thus, the coronary dilatory mechanism of action of levosimendan differs from that of milrinone and it does not seem to be mediated through the activation of PKG. The K+ current through the delayed rectifier K+ channels should be limited during PKG inhibition since these channels are normally activated by PKG [33]. In this situation, the efflux of K+ ions may occur more through the ATP-sensitive K+ channels, which may potentiate the vasodilatory effect mediated through these channels. Recently, patch clamp studies in rat mesenteric arterial cells have revealed that the vasodilation caused by levosimendan may be based on opening of the ATP-sensitive K+ channels [34]. Thus, the interaction of levosimendan and KT5823 on coronary flow could be due to this mechanism of action of levosimendan.

The positive inotropy by levosimendan was found to depend on the strength of electrical stimuli. The phenomenon was not seen with milrinone, indicating that it was not due to PDE III inhibition and cAMP. Furthermore, the β1-adrenoceptor antagonist, atenolol, decreased the sensitivity of twitch tension to the stimulus strength. Thus, the high stimulus strength may have enhanced the release of noradrenaline from adrenergic nerve terminals, which leads to the activation of β1-adrenoceptors and to the direct Gs-protein-mediated opening of calcium channels [35–37]. The high stimulus strength then finally increases intracellular calcium transient, which in turn increases the number of activated cTnC molecules. Since the binding of levosimendan to cTnC is calcium-dependent, it had an increased number of binding sites when more cTnC molecules became activated. Therefore, the calcium sensitization and subsequently the positive inotropy caused by levosimendan was potentiated when a high stimulus strength was used.

Forskolin increases cAMP and the intracellular calcium transient [38]. The increased calcium activates more cTnC molecules and may then potentiate calcium sensitization by levosimendan. In order to minimise this potential artefact in the evaluation of the role of PDE III inhibition and to keep the calcium transient above the critical level, we paced the papillary muscles with the stimulus strength at twofold threshold. The results showed that the positive inotropic effect of milrinone but not that of levosimendan was markedly potentiated by forskolin. Therefore, the role of PDE III inhibition in levosimendan inotropy seems to be negligible. In contrast, PDE III inhibition was an important mechanism for milrinone as suggested in several previous reports [30, 39].

Isoprenaline induces PKA-mediated phosphorylation of cTnI which leads to the altered conformation of cTnC [40, 41]. This may inhibit the binding of levosimendan to cTnC and thereby antagonize its positive inotropy. Our studies with isoprenaline showed that the positive inotropy by levosimendan at the concentration giving a maximum effect was decreased by 60–70% in isoprenaline-pretreated muscles. In contrast, the inotropic response to milrinone was not altered by similar isoprenaline pretreatment. Thus, cTnC seems to be the target protein for levosimendan also in the intact muscle and most of its positive inotropy may then be due to calcium sensitization. Furthermore, when levosimendan was added to the buffer, the positive inotropy by isoprenaline had completely ceased due to its wash-out. Therefore the effect of levosimendan was affected only by the altered phosphorylation of contractile proteins and no longer by the increased intracellular calcium transient. In another study, forskolin and levosimendan were investigated simultaneously, which means that also the increased calcium transient induced by forskolin affected the response to levosimendan. These experimental differences explain why the action of levosimendan was altered differently in the forskolin and isoprenaline studies.

The stepwise increase in levosimendan or milrinone concentrations decreased the positive inotropy by 17 and 45%, respectively, as compared to the effect after a single high concentration. Thus, it seems that the tachyphylaxis developed to milrinone was more marked than that to levosimendan. Furthermore, after washing out of milrinone or isoprenaline the twitch tension decreased below the basal level. This phenomenon of ‘pharmacological stunning’ probably due to the prolonged phosphorylation of cTnI was not observed after washing out of levosimendan.

The calcium sensitization caused by levosimendan occurs at 10 times lower concentrations at 37°C than at 22°C [11]. Therefore, the present skinned fiber experiments carried out at 22°C should not be directly compared with the other studies performed at 37°C. In contrast, our skinned fiber results show how the calcium sensitization by levosimendan was regulated through protein phosphorylation. As indicated by the results, the calcium-sensitizing effect of levosimendan was markedly decreased in the skinned fibers dissected from the isoprenaline-treated heart. Therefore, it is probable that the cAMP-mediated phosphorylation of cTnI [11, 42] antagonises the action of levosimendan on cTnC. This agrees also with the reduced positive inotropy seen at the high PDE III inhibitory concentrations of levosimendan. This phenomenon, however, may in part be due to the opening of ATP-sensitive K+ channels [32]. In contrast, the isoprenaline pretreatment did not antagonize the milrinone inotropy probably because milrinone and isoprenaline change the protein phosphorylation in a similar manner and because milrinone does not act as a calcium-sensitizing agent.

In patients with congestive heart failure (CHF) the sympathetic tone is increased. Despite this, the cAMP-generating pathways and subsequently the phosphorylation of contractile proteins in the failing heart are down-regulated [43]. Therefore, our finding that levosimendan-induced calcium sensitization was antagonised by isoprenaline pretreatment does not mean a lack of effect of levosimendan in CHF. Vice versa, the decreased activity of the cAMP system in the failing heart should favour the calcium-sensitizing effect of levosimendan on contractile proteins.

In conclusion, the positive inotropic effect of levosimendan up to its EC50 concentration was not antagonised by PKA or PKG inhibitors and may therefore be due to the calcium sensitization of contractile proteins. At higher concentrations, the contribution of PDE III inhibition was present and was probably mediated through the increased intracellular calcium transient which potentiates the effect of levosimendan on cTnC. The ratio for calcium sensitization versus PDE III inhibition was roughly 70 versus 30% at the maximum inotropic concentration of levosimendan (0.3 μM). The increased phosphorylation of contractile proteins by PKA antagonised both the calcium-sensitizing and the positive inotropic effect of levosimendan. This supports the hypothesis that calcium sensitization by levosimendan is mediated through cTnC also in intact cardiac muscle. Furthermore, the positive chronotropic effects of levosimendan and milrinone may not be mediated by PKA or PKG. Finally, the mechanism of the coronary-flow-increasing action of levosimendan differs from that of milrinone and is probably not mediated through activation of PKA or PKG. The potentiated effect of levosimendan on coronary flow during PKG inhibition suggests that levosimendan may induce coronary dilatation through the opening of K+ channels.

Acknowledgements

We appreciate the skilful technical contributions of Ritva Huuhilo and Heikki Olkkonen in performing the experiments in the present study.

References

View Abstract